Evaluation of a new complement‐dependent lymphocytotoxicity cross match method using an automated cell counter, the NucleoCounter® NC‐3000™

Complement‐dependent lymphocytotoxicity cross match (CDC‐XM) is the ultimate test of donor/recipient compatibility prior to organ transplantation. This test is based on cell viability, evaluated under fluorescence microscopy by an operator after proper staining. The determination of the positivity threshold may vary depending on the operator. We developed a new method in which the final step of determining cell viability is automated using the NC‐3000™ (Chemometec®), an image cytometer able to precisely determine the percentage of dead/live cells in a suspension. After T and B donor cells isolation by negative selection, complement‐dependent lysis was performed in macrovolumes in a PCR plate. Then, cell viability was measured by the NC‐3000™. The sensitivity and routine CDC‐XM results of this new method were compared to those of CDC‐XM reference method using Terasaki plates. The sensitivity of CDC‐XM expressed in the ASHI scoring system of this method was similar to the reference method results for a dilution range of the positive controls. Similarly, the results of the new method were comparable in a clinical situation to those obtained with the reference method after a study of 10 cross‐matches, of which 5 cross‐matches with DSA were positive and five cross‐matches without DSA were negative. Moreover, ASHI scores were similar to those obtained using the reference method, and the mortality percentage was reproducible (CV < 15%). The assessment of cell viability by the NC‐3000™ is easy to perform and highly reproducible but requires CDC‐XM to be performed by the macrovolume method. The determination of a precise percentage of viability/mortality by the automation excludes operator variability and allows a better understanding of results close to the decision threshold.


K E Y W O R D S
CDC cross match, NC-3000, transplantation

| INTRODUCTION
The complement-dependent lymphocytotoxicity cross match (CDC-XM) is the ultimate compatibility test between an organ donor and a recipient. Today, with the exception of kidneys transplanted in a hyperimmunized patient, it is carried out retrospectively, that is, after the transplant within a maximum regulatory period of 48 h. The transplant decision is made based on the result of the virtual cross match (vXM), 1,2 which reduces the duration of cold ischemia of the graft 3 by avoiding carrying out the CDC-XM, which can take up to 4 h. However, the interpretation of the vXM results is complex, requires considerable expertise and is difficult to standardize from one laboratory to another. 4 The CDC-XM therefore retains its place in the biological management of the transplanted patient at the present time. 5 However, CDC-XM is actually being used less and less and not at all in many transplantation centers, due to its complex interpretation and its relative lack of sensitivity to identify HLA-DSA present at the time of transplant. It is also accepted that flow cytometry crossmatch (FC-XM) should be used to select donor instead to CDC-XM. 6 Indeed, FC-XM is more sensitive and seems to detect a higher risk of inferior allograft outcome, even in negative result of CDC-XM. 6,7 However, in many centers, FC-XM assessment is not performed in the on call period and its DSA risk stratification is variable among the centers.
The CDC-XM is well known in principle. It was miniaturized thanks to the work of Terasaki in 1964 and is produced in microplates in a drop of oil that serves as a microreactor. 8 Briefly, the donor's T and B lymphocytes are isolated and brought into contact with the recipient's serum in the presence of rabbit complement. After an appropriate incubation time allowing cell sensitization and complement-dependent lysis, the percentage of cell death is determined using vital staining by counting live cells and dead cells under a fluorescence microscope. The use of serum treated with DTT and serum not treated with DTT makes it possible to determine the nature of the antibody present (IgG vs. IgM). 9 The reading step of the CDC-XM is performed by an authorized technician. This reading is subject to interindividual variability since it involves the human eye. In addition, the CDC-XM is an on-call technique, performed in an emergency situation, often at night: the stress and fatigue of the operator can add to this variability. It is therefore of interest to be able to automate the reading step.
Thus, we propose to develop a new XM method using an automated cell counter, the NucleoCounter ® NC-3000™, developed by the company Chemometec. The NC-3000™ is an image cytometer that uses fluorescence imaging technology. Cell suspensions are labeled with a solution composed of Acridine Orange (AO) and DAPI. AO is a DNA intercalating agent that penetrates the cell membrane and binds to DNA in living or dead cells, whereas DAPI, at the concentrations used, does not have the ability to penetrate living cells and binds the double-stranded DNA of dead cells only. 10 Therefore, after a quick incubation, the live cells are AO positive, while the dead cells are positive for AO and DAPI. The number of positive AO events alone and the number of positive AO and DAPI events are counted to determine the cell quantity (expressed in cells per milliliter) in the suspension and cell viability/ mortality. However, this device needs a volume that is not suitable on a Terasaki plate to count cells. Thus, a CDC-XM macromethod has been developed. The aim of this study is to secure the critical reading step of the CDC-XM by proposing an alternative and automated protocol.

| Samples
Blood donations preserved with EDTA from five healthy donors were collected in the "Etablissement Francais du Sang," in accordance with BSL-2 practices. A medical interview was carried out prior to blood donation to exclude donors with medical contraindications. This study was carried out in accordance with the French Public Health Code (art L1221-1), approved by the institutional ethics committee and conducted in compliance with the Good Clinical Practice Guidelines and the Declaration of Helsinki and Istanbul. For routine CDC-XM, the piece of spleen received in the laboratory was placed in a Petri dish with 4 mL of cold saline.
The donor's lack of opposition to the scientific research was verified. The spleen was perforated and perfused several times with a syringe with a 1.1 mm diameter needle. The cell suspension obtained was collected.

| Preparation of the sera
The serum sample collected that day and a historical serum sample, if available, of the patient were brought to room temperature. For each serum sample, pretreatment with Dithiothreitol (DTT) at a 1:10 dilution was performed to destroy any IgM. The aliquots were then incubated in an oven at 37 C for 15 min to allow the action of DTT. The positive control for T cells was a monoclonal serum containing anti-total lymphocyte IgG capable of binding complement (Anti-Lymphocyte IgG, ALSG, One-Lambda, Thermo Fisher, Waltham, MA, USA). The positive control for B cells was a monoclonal serum containing complement-fixing anti-B-cell IgG (Anti-Lymphocytes B, ABSG, OneLambda, Thermo Fisher, Waltham, MA, USA). The negative control was a patient's serum certified by the single-antigen technique (Labscreen Single Antigen class I and II, Thermo Fisher, Waltham, MA, USA) to be free of anti-HLA antibodies.

| Isolation of donor T and B lymphocytes by negative selection
T and B lymphocytes were extracted from the cell suspension using EasySep™ kits from STEMCELL, following the manufacturer's protocol. The EasySep™ Direct HLA T-Cell Isolation Kit and EasySep™ Direct HLA B-Cell Isolation Kit were used for T and B lymphocyte isolation, respectively.
For T cells, 2 mL of cell suspension was collected in a 14 mL (17 Â 95 mm) round-bottom tube. Then, 100 μL of EasySep™ Direct HLA T-Cell Isolation cocktail (50 μL per mL of suspension) was added. After mixing by aspiration and dispensing, the sample was incubated for 5 min on the benchtop at room temperature to allow antibody action. EasySep™ Direct RapidSpheres™ beads were vortexed for 30 s, and 100 μL of the bead suspension was added (same volume as the isolation cocktail volume). Two milliliters of saline were added, and the resulting diluted sample was mixed and then placed on a STEMCELL "The Big Easy" magnet for 3 min. Without removing the tube from the magnet, by simple inversion of the magnet, the content of the first tube was transferred into a second identical tube (tube 2). An additional 100 μL of the EasySep™ Direct RapidSpheres™ bead suspension was added. After mixing by aspiration and dispensing, the sample was incubated for 3 min on the benchtop, and then tube 2 was placed on the magnet for 3 min. Without removing the tube from the magnet, by simple inversion, the purified suspension of T lymphocytes was transferred into a collection tube.
The same process, with the same volumes and incubation times, was performed for the isolation of B lymphocytes using the adapted kit and the EasySep™ Direct HLA B-Cell Isolation cocktail.

| Isolation of donor T and B lymphocytes by positive selection
Isolation of total lymphocytes was performed using the Ficoll method. The quantity of total lymphocytes and cell viability were determined using Trypan blue in Malassez cells. If the count was less than 5 Â 10 6 and/or the viability was less than 80%, then a new isolation was performed. The suspension was divided into two labeled 15 mL tubes. For the T lymphocytes, Dynabeads HLA class I beads (Invitrogen by Thermo Fisher), coated with an anti-CD8 antibody, were used. The beads were vortexed for 30 s, and 100 μL was added to the lymphocyte suspension. After 5 min on a rotary agitator, the tubes were then placed on a magnet for 2 min. The supernatant, corresponding to the non-T-cell fraction, was removed. The beads were resuspended in 5 mL of cold saline solution and placed on a magnet for 2 min. After discarding the supernatant, the beads were taken up in 400 μL of RPMI 10% FCS. The same protocol was applied in parallel for the B lymphocytes using Dynabeads HLA class II beads (Invitrogen by Thermo Fisher) coated with a pan anti-HLA DR antibody.

| Concentration and calibration of the cell suspension
The isolated lymphocyte count was performed on the NC-3000™ using the Via-1 Cassette™ device. The Via-1 Cassette aspirated a 60 μL volume of suspension and automatically stained it with a calibrated volume of AO and DAPI. The device was then placed in the NC-3000™, which determined the cell concentration of the suspension from a count of the cells present in the 1.4 μL reading chamber. A graph representing the cells in the form of a scatter plot displays the surface of AO in the ordinate and the intensity of the spot in the abscissa, with each spot corresponding to a cell.
After centrifugation of the cell suspension, a volume of supernatant was added to obtain a cell concentration of approximately 1.10^6 cells/mL depending on the concentration of the initial suspension. Targeting such a concentration allowed us to calibrate our suspension within the optimal reading range of the NC-3000™ instrument (5 Â 10 4 to 5 Â 10 6 cells/mL), since the suspension will be further diluted by the addition of serum and complement in the remainder of the protocol.

| CDC-XM method using Terasaki plates
The CDC-XM micromethod was performed using oiled Terasaki plates. The distribution was performed in duplicate for 2, 1, and 1 μL at 1:3 of the untreated and treated patient sera/DTT and positive and negative controls B lymphocytes against 1 μL of T, B, non-B and non-T lymphocytes according to the distribution scheme ( Figure 1). After a 30-min incubation, 2-6 μL of rabbit complement was added to each well during a one-hour incubation. Finally, 4 μL of FluoroQuench was added to each well. After a 10-min incubation, the reading was then carried out on a fluorescence microscope. A score was assigned to each well according to the percentage of mortality evaluated by eye according to the ASHI standards.

| CDC-XM macromethod
The distribution scheme is described in Figure 1B.
PCR-96-FLT-C 96-well microplates from Axygen were used. According to the plate design, 5 μL of cell suspension and 5 μL of DTT-treated or untreated serum were distributed in the wells, as well as the positive and negative controls. After a brief centrifugation, the plates were incubated for 30 min on a plate shaker at 600 rpm. After this first incubation, 18 μL of rabbit complement was added to each well. After another brief centrifugation, the plates were incubated for 60 min on a plate shaker (Grant-bio) at 600 rpm. After this second incubation, 2 μL of Chemometec Solution 13, a mix of AO (30 μg/mL) and DAPI (100 μg/mL), was added to each well. After a brief centrifugation, the plates were incubated for 2 min on a plate shaker at 600 rpm. The cell suspensions were then ready for reading on the NC-3000™.

| Viability measurement on the NC-3000™
After homogenization, 10 μL was dispensed into a calibrated reading chamber on a glass slide (NC-Slide A8, Chemometec). Since the final volume in each well was 30 μL, the same well was able to be analyzed several times to increase the reliability of the result. Each slide contained eight chambers and therefore allowed 8 tests compatible with the NC-3000™ automaton. Using the computer software NucleoView NC-3000™ (Version 2.1.25.12), each chamber of the slide was identified, and the volume of solution 13 was filled in to allow the instrument to take into account the induced dilution factor. The reading protocol was selected (default protocol, "Viability and cell count assay"). The slide was then placed into the device, and the reading was launched.
The reading time per slide was approximately 3 min. The device took three pictures per chamber at two fluorescence wavelengths to reveal the staining of the cells first by AO (green fluorescence) and then by DAPI (blue fluorescence). Total cells correspond to AO-positive cells. Dead cells correspond to cells positive for both AO and DAPI. Firstly, the software displayed the raw data of cell count (cells/milliliter) and the percentage of viability/mortality. The "Plot Manager" tool integrated in the software allowed us to analyze the data in more detail by representing the cells as scatter plots: total cells (AO area versus AO intensity) and dead cells (DAPI area versus DAPI intensity). On these graphs, the gating was manually changed to better define the cell population being analyzed by excluding debris. Finally, the number of total cells (per milliliter of suspension), the number of live cells, the number of dead cells, the percentage of live/ dead cells and the associated ASHI score were produced by Chemometec software in the Excel files ( Figure 2).

| Statistical Analysis
The results obtained in terms of ASHI score were compared to those obtained by manual reading from the CDC-XM reference technique. The precise percentage of mortality from the device allowed for a more accurate interpretation than that in the CDC-XM, which did not provide this information. The percentages of mortality were rounded up to the next whole number. For positive control dilutions, each condition (T lymphocytes + DTT, T lymphocytes without DTT, B lymphocytes + DTT, and B lymphocytes without DTT) was tested twice per plate, and each plate was run in triplicate for each dilution.
Thus, the coefficient of variation (CV) of the cell mortality measurement (mean/standard deviation*100) was determined for T lymphocytes and B lymphocytes with or without DTT treatment from 6 measurements. On each line of each plate, a positive control and a negative control were used, which enabled us to determine the CV of the positive and negative controls for T and B lymphocytes from three measurements for each dilution. For the negative and positive CDC-XM, the same plate design was used, but in addition, each well was analyzed twice, allowing the calculation of CVs for each cross match and each condition over a total of 12 measurements and over 6 measurements per cross match for positive and negative controls.

| Macromethod evaluation
The CDC macromethod was first evaluated by comparing the results of the macromethod with those of the F I G U R E 1 A. Layout of the plate in the reference method; B. Layout of the plate in the NC-3000™ method.
reference method using dilutions of the positive control (Table 1). In the CDC macromethod, all nondiluted positive controls were positive, and all negative controls were negative. The results of the positive control dilutions using the macromethod were correlated with those obtained using the reference method. For the T lymphocytes, the two methods resulted in a score of 6 in the 1:8 dilution, a score of 4 in the 1:16 dilution and a negative score for the subsequent dilutions. It should be noted that for the 1:32 and 1:64 dilutions, the absolute value of mortality was at the limit between a score of 2 and a score of 4, but taking into account the basal  1 mortality allowed us to determine a score of 2, corresponding to a negative cross match. For the B lymphocytes, the two methods resulted in a score of 8 for the 1:8, 1:16, and 1:32 dilutions, a score of 6 for the 1:64 dilution and finally a negative score for the subsequent dilutions. As few HLA laboratories use positive lymphocyte isolation to perform the CDC-XM, we compared the performance of the two lymphocyte isolation methods from the spleen in the CDC-XM macromethod ( Table 2). The cell mortality of the negative control was higher in positive isolation than in negative isolation (15% vs. 4.7% for T lymphocytes and 16% vs. 2.3% B lymphocytes). The positive control results were not different as a function of the type of cell isolation. The negative serum control produced the highest value of mortality for T and B lymphocyte positive isolation compared to negative isolation. We first note that the scores obtained by negative sorting correspond to an unequivocal negative. A significant toxicity of DTT on B lymphocytes from positive isolation was detected by the NC-3000™. The average CV of the values was not significantly different in positive sorting and in negative sorting. However, in the macromethod, negative sorting seemed better correlated with the reference method and was more precise due to lower basal mortality.

| Macromethod sensitivity and specificity evaluation
We evaluated the accuracy of the CDC macromethod results. The values of the positive control were all >95%. The average CV of the positive controls for all the CDC macromethods performed was very low (1.1%). The CV of the negative controls for the CDC macromethods was higher but remained lower than 30% (25.6%).

| Comparison of two methods in clinical condition
The macromethod was tested in parallel with the reference method on 5 CDC-XM-positive performed artificially from spleen lymphocytes of known HLA types and sera presenting donor-specific antibodies (DSAs) with high MFI and directed against HLA-A, HLA-B, HLA-DR, or HLA-DP (Table 3). Sera 1 and 5 were positive on T and B cells without ambiguity and sera 2 to 4 were positive only on B cells without ambiguity. All ASHI scores obtained with the NC-3000™ were correlated with those obtained with the reference technique, except Serum 4 on the untreated B lymphocytes, where the score of the NC-3000 was higher than the score of the reference method without modifying the CDC-XM result. The average CV of mortality reported by the NC-3000™ was 9.6% for CDC-XM-positive patients.
Finally, five CDC-XM tests were prospectively performed by macromethod and reference method (Table 4). In the two methods, all results were identical and were negative for T and B lymphocytes, as expected. The average CV of the CDC-XM analytical results from the NC-3000 was 13.5% for the CDC-XM negative control. Interestingly, the comparison of analytical XM results with the negative control allowed optimization of the negativity of results, independent of the ASHI score.

| DISCUSSION
This study allowed us to determine whether the NC-3000™ (NucleoCounter ® ) could be used as an automatic cell reader in the CDC-XM protocol. However, this automation required a new CDC-XM method. Indeed, the final reaction needed to be modified to obtain an optimal cell count; the final volume in the Terasaki plate is too low, and pipeting the cell suspension through the drop of oil before the reading step is very difficult. Different final volumes for our reaction medium were tested; proportionally increasing the volumes of the micromethod made it possible to adhere as closely as possible to the reference method. A volume that was too low did not always allow the reading chamber of the NC-3000™ to be filled due to evaporation. A volume that was too high consumed reagents unnecessarily. The final volume of the protocol that we propose is a compromise that also makes it possible to perform two analyses on the same well to compensate for one failure or to increase the reliability of the analysis and obtain accurate data.
The macromethod is performed in a PCR-type microplate without oil, which constitutes a notable change. A plate with a conical bottom well was selected because evaporation is limited and the contact between the different reagents is optimal. In addition, this plate model is available in all HLA Laboratories. To optimize the contact between the different reagents and therefore the cell sensitization and complement-dependent lysis, in the absence of a drop of oil, incubations with shaking at 600 rpm were performed. This agitation is necessary to promote the mixing of the reagents that are added one by one to the wells and to avoid sedimentation of the cells over time.
The macromethod was then evaluated using dilutions of the positive control and by comparing the results of the macromethod with those of the reference method. The macromethod therefore has the same sensitivity as the reference micromethod over the positive control dilution range.
In the macromethod, DTT sometimes increases the mortality of the sorted cells but without major impact on the score and with a similar intensity to that with the reference method. 11 Two lymphocyte isolation methods were tested: positive sorting and negative sorting. The isolation of T and B lymphocytes by positive sorting led to greater cell death than negative sorting at the end of the lymphocyte isolation step, when the two isolations were performed in parallel from the same initial blood sample. In the reference method, the two isolation techniques are used according to the local HLA laboratory practices 12,13 and give concordant and reliable results. It should be noted that negative sorting, which is less traumatic for the cells, is deemed to have a slightly lower sensitivity than positive sorting, which predisposes the cells to lysis. This predisposition may be linked to the mechanical effect of the beads, which causes shocks in the suspension, particularly during homogenization and the activation of stimulation systems by the bead-antibody-cell bond. The higher basal mortality in positive sorting may make it more difficult for the human eye to read the CDC-XM under a fluorescence microscope. In the macromethod, only negative sorting should be used. Indeed, the beads cause very quick sedimentation of the cells at the bottom of the wells during the incubation step, allowing for a failure of cell sensitization and complement-dependent lysis. Furthermore, when filling the reading chambers of the A8 slides, the distribution of the beads was not perfectly homogeneous. The analysis of the pictures of the slides showed that the cells from the positive sorting were grouped together in clusters, and the automation differentiated each cell with difficulty. Thus, a group of several cells could be counted as a single larger cell. Conversely, the cells from negative sorting were well individualized and formed a homogeneous layer that the automation could analyze with precision.
Artificial CDC-XM positivity showed that the macromethod had the same sensitivity as the reference method on lymphocytes from the spleen. The average coefficients of variation of mortality reported by the NC-3000™ were less than 15%, which corresponds to good analytical performance.
Finally, the macromethod was prospectively tested in parallel with the reference method on CDC-XM. The results obtained, which were all negative as expected, were well correlated with the reference method. However, an interpretation step of comparison with the basal mortality of the negative control was necessary in a few cases. The mortality of cells extracted from spleens was often high. Compared to the NC-3000™, the cell mortality was underestimated by the technicians in the reference technique. Sera 1 to 4 were considered negative because the percentage of dead cells, although high, was identical to that of the negative control. The result by automation was therefore more precise, and its interpretation did not require an operator with much experience with the method. This is very interesting because the CDC-XM is performed less often in the laboratory, 14 and a loss of technical experience will necessarily take place in the years to come. In addition, the CDC-XM is often performed at night and under stress. The results obtained by automation are less subject to a variation in interpretation related to the operator's level of fatigue, stress or experience.
A layout of the plate that makes it possible to carry out the cross match on only 3 A8 blades has been determined. According to this plate layout, the serum of the patient collected that day and treated or not treated with DTT can be placed in duplicate. Historical treated and untreated sera were also tested. Each slide had its own negative control for T and B lymphocytes. Finally, the effectiveness of DTT was checked using an IgM control treated and not treated with DTT. This plate design provides the same information as the reference technique by limiting the number of wells to be prepared as much as possible and therefore limiting the cost and time associated with the analysis. Including cell isolation, which is the same process for both methods, the technical time was estimated at 45 min for the macromethod and at 35 min for the reference method. The incubation times are identical, but it should be noted that the reading is performed after 2 min of incubation with solution 13 in the macromethod and after 10 min of exposure to Fluoro-Quench in the reference method. Interestingly, the macromethod does not consume more reagents than the reference method used in our HLA Laboratory. This is mainly due to the number of wells, which is lower in the macromethod.
Finally, our study shows that the results of CDC-XM using the macromethod with automated reading by the NC-3000™ are reproducible and accurate and have an identical sensitivity to the reference technique. The method works for negative lymphocyte isolation from spleen or whole blood, so it can be applied to all clinical situations in which an CDC-XM is needed, including in the context of living donation. The macromethod requires more technical time but has no impact on the regulatory rendering time of 4 h. It is not more expensive if the acquisition of the automation equipment is excluded. It is even slightly more economical in terms of reagents. This method makes it possible to remove the subjectivity implied by reading with the human eye in the analysis and does not require a great amount of technical experience. It is therefore more secure and better adapted to the on-call situations in which the